Research On Explosives Detection By Electronic Nose

This article mainly focuses on the metal oxide gas sensor arrays and their applications on the detection of explosives.Firstly ZnO nanoparticals were produced by renovated hybrid induction and laser heating,then fabricated thick film gas sensors.We use dynamic sampling method to choose gas sensors whose sensitivities and stabilities are best to the fingerprint gases of picric acid, NH4NO3, DNT and mineral explosive.Then we use stastic sampling method to exam the detection ability of the optimized sensor arrays.The results show that the detection concentration of the four explosives are all low to 3.34μg/L and with the increase of concentration of analytes, the sensitivities of the sensors increased as an exponential function. In order to approach the practical application on detection of solid explosives by electronic noses, we used mass flowmeter to control the concentration of the fingerprint which entered into the test chamber to react with the sensor arrays for the sake of detecting the lest concentration of explosives. The results of the discrimination of the four explosives at different concentration are as follows:At 4.0mg,using different feature parameters to Principal Component Analysis(PCA),Cluster Analysis(CA)and Discriminant Function Analysis(DFA),four explosives can fully be recognized.At 83.3μg/L and 15.4μg/L,using appropriate feature parameters to DFA analysis, four explosives can fully be recognized too.At 3.34μg/L,using RBF analysis,the error rate is only 3.12%.At last the humidity disturbing experiment is to test the change of the air resistances and maximum (minimum) resistances of the sensor arrays at different environment humidities. And the results show humidity is an important effect factor on sensor's use. Testing the base-line resistances of the six sensors in air within a range of humidity revealed that the humidity has less effect on the resistances of the six picked sensors within the range of 75-97RH%, therefore, the specific testing condition is very important to ensure the reliability of analyse result.The paper divide into four parts: the first part describes the development and system construction of electronic nose, investigation of explosives detection and the content, means, and technical route of our task focoused on.The second part introduce the experiment equipment of electronic nose and the instruction of each experiment.The third part presentate the data processing including pretreatment of data and feature extraction, pattern recognization.The final part is the declaration of the results discussion and analysis.